Temperature curve
The temperature curves of a GPU differ significantly between constant stress tests and dynamic gaming scenarios, which is due to the different load profiles and the use of modern power management technologies. In a stress test, the GPU is operated with a constant workload throughout. The temperature curve is comparatively smooth and stable here, as the energy consumption and heat development are almost constant over the entire duration. In this scenario, all functional areas of the GPU are continuously active and mechanisms such as power gating are hardly used. The heat development remains stable, as the energy supply is even and the cooling works at a constant level. This is reflected in a linear temperature development in which the GPU efficiently utilizes its thermal limits without being additionally burdened by frequent fluctuations.
In gaming scenarios, on the other hand, the temperature curve is characterized by dynamic fluctuations. The reason for this lies in the strongly changing performance requirements caused by the different GPU calculations. Some frames require intensive calculations, for example for complex shaders or ray tracing scenes, while others can be processed with comparatively few resources. Technologies such as power gating ensure that unused GPU areas are selectively deactivated, which significantly reduces heat development during periods of low load. At the same time, separate power rails ensure the constant supply of critical components such as the memory, even if the GPU cores are temporarily turned down.
Another decisive factor is Accelerated Frequency Switching, which enables rapid adjustments to the clock frequency and voltage. This technology allows the GPU to react to load changes in microseconds and adjust its performance accordingly. This optimizes energy consumption, but at the cost of a more irregular temperature curve. Particularly in scenarios in which the GPU switches between low and high clock frequencies several times within a short period of time, visible temperature fluctuations occur. The heat development follows the dynamic changes in power consumption, which is a challenge for the cooling system as it has to constantly react to the varying thermal conditions.
The memory temperatures of the MSI RTX 5090 SUPRIM vary significantly between silent and gaming mode, which is due to the different performance and cooling profiles of the two modes.
The cooling of the memory modules is supported by the large vapor chamber and the high-quality thermal pads. In silent mode, the lower fan speeds affect the cooling performance, but this is compensated for by the overall lower waste heat. In Gaming Mode, on the other hand, the fan profile is more aggressive, which controls the higher waste heat from the VRAM. The efficiency of heat dissipation also depends on the quality of the thermal pads used (see material test).
Clock rates
The clock rates of the MSI RTX 5090 SUPRIM differ significantly between Silent Mode and Gaming Mode, which is due to the different TDP limitations and fan profiles of the two modes. These differences have a direct impact on the performance and thermal load of the card. In silent mode, the clock rates are more moderate, as the maximum TDP is limited and the card is designed to minimize energy consumption and noise. Under typical load, the boost clocks range from around 2,600 to 2,700 MHz, depending on the thermal situation and the performance requirements of the game. The lower clock rates ensure that the GPU generates less heat, which enables quieter cooling and more stable temperatures. However, the performance remains somewhat limited due to the lower clock rate compared to gaming mode, which is reflected in a slight drop in average frame rates.
In gaming mode, on the other hand, the card is designed for maximum performance and raises the TDP to up to 600 watts. This allows significantly higher clock rates, which can reach peaks of almost 2,800 MHz under load. The increased clock frequency leads to a noticeable increase in performance in graphics-intensive applications, especially at high resolutions such as QHD and 4K. However, the higher clock rates also increase heat generation and fan speeds, which leads to louder operation. Energy consumption increases accordingly, as the GPU draws more power due to the higher clock rates.
Thermography
The thermographic images of the MSI RTX 5090 SUPRIM after 30 minutes each in idle, gaming and torture mode clearly show the thermal distribution and efficiency of the cooling design. The infrared measurements with the Optris PI 640 illustrate how the cooling works under different loads, with both hotspots and temperature areas with even heat dissipation becoming visible in detail. The thermography shows that the MSI RTX 5090 SUPRIM has a very efficient cooling design that ensures stable temperatures both under moderate load in gaming and under extreme load in the Torture test. The “Double Flow Through” design and the high-quality thermal materials make a significant contribution to protecting critical components from overheating and ensuring even thermal distribution.
- 1 - Introduction, overview and technical data
- 2 - Test system and equipment
- 3 - Teardown: PCB, components and cooler
- 4 - Material analysis and heat conducting materials
- 5 - Gaming performance
- 6 - Power consumption, load peaks, PSU recommendation
- 7 - Temperatures, clock rates and thermography
- 8 - Fan curves and operating noise
- 9 - Summary and conclusion













































208 Antworten
Kommentar
Lade neue Kommentare
Veteran
Neuling
Neuling
Mitglied
Veteran
Mitglied
Mitglied
Veteran
Urgestein
Urgestein
Mitglied
Veteran
Urgestein
1
1
Mitglied
Veteran
Veteran
Veteran
Alle Kommentare lesen unter igor´sLAB Community →